Event-Related Potential
Total Page:16
File Type:pdf, Size:1020Kb
Event-related potential techniques. In 1935–1936, Pauline and Hallowell Davis recorded the first known ERPs on awake humans and their findings were published a few years later, in 1939. Due to World War II not much research was conducted in the 1940s, but research focusing on sensory issues picked back up again in the 1950s. In 1964, research by Grey Walter and colleagues began the modern era of ERP component discoveries when they reported the first cognitive ERP component, called the contingent nega- tive variation (CNV).[3] Sutton, Braren, and Zubin (1965) made another advancement with the discovery of the P3 component.[4] Over the next fifteen years, ERP compo- nent research became increasingly popular. The 1980s, with the introduction of inexpensive computers, opened up a new door for cognitive neuroscience research. Cur- A waveform showing several ERP components, including the rently, ERP is one of the most widely used methods in N100 and P300. Note that the ERP is plotted with negative volt- cognitive neuroscience research to study the physiological ages upward, a common, but not universal, practice in ERP re- correlates of sensory, perceptual and cognitive activity as- search sociated with processing information.[5] An event-related potential (ERP) is the measured brain response that is the direct result of a specific sensory, cognitive, or motor event.[1] More formally, it is any 2 Calculation stereotyped electrophysiological response to a stimulus. The study of the brain in this way provides a noninvasive means of evaluating brain functioning in patients with ERPs can be reliably measured using cognitive diseases. electroencephalography (EEG), a procedure that measures electrical activity of the brain over time ERPs are measured by means of electroencephalography using electrodes placed on the scalp. The EEG reflects (EEG). The magnetoencephalography (MEG) equivalent thousands of simultaneously ongoing brain processes. of ERP is the ERF, or event-related field.[2] This means that the brain response to a single stimulus or event of interest is not usually visible in the EEG recording of a single trial. To see the brain’s response to 1 History a stimulus, the experimenter must conduct many trials and average the results together, causing random brain activity to be averaged out and the relevant waveform to With the discovery of the electroencephalogram (EEG) [6] in 1924, Hans Berger revealed that one could measure remain, called the ERP. the electrical activity of the human brain by placing elec- The random (background) brain activity together with trodes on the scalp and amplifying the signal. Changes other bio-signals (e.g., EOG, EMG, EKG) and electro- in voltage can then be plotted over a period of time. He magnetic interference (e.g., line noise, fluorescent lamps) observed that the voltages could be influenced by exter- constitute the noise contribution to the recorded ERP. nal events that stimulated the senses. The EEG proved to This noise obscures the signal of interest, which is the be a useful source in recording brain activity over the en- sequence of underlying ERPs under study. From an en- suing decades. However, it tended to be very difficult to gineering point of view it is possible to define the signal- assess the highly specific neural process that are the focus to-noise ratio (SNR) of the recorded ERPs. The reason of cognitive neuroscience because using pure EEG data that averaging increases the SNR of the recorded ERPs made it difficult to isolate individual neurocognitive pro- (making them discernible and allowing for their interpre- cesses. Event-related potentials (ERPs) offered a more tation) has a simple mathematical explanation provided sophisticated method of extracting more specific sensory, that some simplifying assumptions are made. These as- cognitive, and motor events by using simple averaging sumptions are: 1 2 4 ADVANTAGES AND DISADVANTAGES 1. The signal of interest is made of a sequence of event- a number indicating either the latency in milliseconds or locked ERPs with invariable latency and shape the component’s ordinal position in the waveform. For instance, a negative-going peak that is the first substantial 2. The noise can be approximated by a zero-mean peak in the waveform and often occurs about 100 mil- 2 Gaussian random process of variance σ which is liseconds after a stimulus is presented is often called the uncorrelated between trials and not time-locked to N100 (indicating its latency is 100 ms after the stimulus the event (this assumption can be easily violated, for and that it is negative) or N1 (indicating that it is the first example in the case of a subject doing little tongue peak and is negative); it is often followed by a positive movements while mentally counting the targets in an peak, usually called the P200 or P2. The stated latencies oddball paradigm). for ERP components are often quite variable. For exam- ple, the P300 component may exhibit a peak anywhere Having defined k , the trial number, and t , the time between 250ms – 700ms.[9] elapsed after the k th event, each recorded trial can be written as x(t; k) = s(t) + n(t; k) where s(t) is the sig- nal and n(t; k) is the noise (Note that, under the assump- tions above, the signal does not depend on the specific 4 Advantages and disadvantages trial while the noise does). The average of N trials is 4.1 Relative to behavioral measures Compared with behavioral procedures, ERPs provide a 1 XN 1 XN x¯(t) = x(t; k) = s(t) + n(t; k) continuous measure of processing between a stimulus N N and a response, making it possible to determine which k=1 k=1 stage(s) are being affected by a specific experimental ma- The expected value of x¯(t) is (as hoped) the signal itself, nipulation. Another advantage over behavioral measures E[¯x(t)] = s(t) . is that they can provide a measure of processing of stim- uli even when there is no behavioral change. However, Its variance is because of the significantly small size of an ERP, it usu- ally takes a large number of trials to accurately measure 2 it! correctly.3 [10] h i N 2 N 1 X 1 X [ ] σ2 Var[¯x(t)] = E (¯x(t) − E[¯x(t)])2 = E 4 n(t; k) 5 = E n(t; k)2 = N 2 N 2 N k=1 k=1 4.2 Relative to other neurophysiological For this reasonp the noise amplitude of the average of N measures trials is 1/ N times that of a single trial. Wide amplitude noise (such as eye blinks or movement 4.2.1 Invasiveness artifacts) are often several orders of magnitude larger than the underlying ERPs. Therefore, trials containing Unlike microelectrodes, which require an electrode to be such artifacts should be removed before averaging. Ar- inserted into the brain, and PET scans that expose hu- tifact rejection can be performed manually by visual in- mans to radiation, ERPs use EEG, a non-invasive proce- spection or using an automated procedure based on pre- dure. defined fixed thresholds (limiting the maximum EEG am- plitude or slope) or on time-varying thresholds derived from the statistics of the set of trials.[7] 4.2.2 Spatial and temporal resolution 3 Nomenclature of ERP compo- ERPs provide excellent temporal resolution—as the speed of ERP recording is only constrained by the sam- nents pling rate that the recording equipment can feasibly sup- port, whereas hemodynamic measures (such as fMRI, ERP waveforms consist of a series of positive and neg- PET, and fNIRS) are inherently limited by the slow speed ative voltage deflections, which are related to a set of of the BOLD response. The spatial resolution of an underlying components.[8] Though some ERP compo- ERP, however, is much poorer than that of hemody- nents are referred to with acronyms (e.g., contingent neg- namic methods—in fact, the location of ERP sources is ative variation – CNV, error-related negativity – ERN, an inverse problem that cannot be exactly solved, only es- early left anterior negativity – ELAN, closure positive timated. Thus, ERPs are well suited to research questions shift – CPS), most components are referred to by a letter about the speed of neural activity, and are less well suited (N/P) indicating polarity (negative/positive), followed by to research questions about the location of such activity.[1] 3 4.3 Cost staring at the grid, the subject may communicate which stimulus he is looking at, and thus slowly “type” words.[17] ERP research is much cheaper to do than other imaging Other ERPs used frequently in research, especially techniques such as fMRI, PET, and MEG. This is be- neurolinguistics research, include the ELAN, the N400, cause purchasing and maintaining an EEG system is less and the P600/SPS. expensive than the other systems. 5 Clinical ERP 7 ERP software and training re- sources Physicians and neurologists will sometimes use a flashing visual checkerboard stimulus to test for any damage or • EEGLAB Toolbox – A freely available, open- trauma in the visual system. In a healthy person, this stim- source, Matlab toolbox for processing and analyzing ulus will elicit a strong response over the primary visual EEG data cortex located in the occipital lobe, in the back of the brain. • ERPLAB Toolbox – A freely available, open- ERP component abnormalities in clinical research have source, Matlab toolbox for processing and analyzing been shown in neurological conditions such as: ERP data • dementia[11] • The ERP Boot Camp – A series of training work- shops for ERP researchers • Parkinson’s disease[12] • multiple sclerosis[13] 8 See also • head injuries[14] [15] • stroke • Bereitschaftspotential • obsessive-compulsive disorder[16] • Brain vital signs • 6 Research ERP C1 and P1 • Contingent negative variation ERPs are used extensively in neuroscience, cognitive psy- chology, cognitive science, and psycho-physiological re- • Difference due to memory search.